Extending the FIP bias sample to magnetically active stars. Challenging the FIP bias paradigm

The different elemental abundances of the photosphere and the corona are striking features of not only the Sun, but other stars as well. This phenomenon is known as the FIP effect (FIP stands for first ionization potential), and its strength can be characterized by the FIP bias, the logarithmic abundance difference between low- and high-FIP elements in the corona, compared to the photosphere. The FIP bias was shown to depend on the surface temperature of the star. We compiled FIP bias and other parameters for 59 stars for which coronal composition is available, now including evolved stars. Using principal component analysis and linear discriminant analysis, we searched for correlations with other astrophysical parameters within the sample which may influence the stellar FIP bias. Adding stars to the $T_{\rm eff}-$FIP bias diagram unveiled new features in its structure. In addition to the previously known relationship, there appears to be a second branch, a parallel sequence about 0.5 dex above it. While the $T_{\rm eff}$ remains the main determinant of the FIP bias, other parameters such as stellar activity indicators also have influence. We find three clusters in the FIP bias determinant parameter space. One distinct group is formed by the evolved stars. Two groups contain main sequence stars in continuation separated roughly by the sign change of the FIP-bias value. The new branch of the $T_{\rm eff}-$FIP bias diagram contains stars with higher activity level, in terms of X-ray flux and rotational velocity. The two main sequence clusters run from the earliest spectral types of A-F with shallow convection zones through G-K-early M stars with gradually deeper convection zones, and end up with the fully convective M dwarf stars, depicting the change of the dynamo type with the internal differences of the main sequence stars in connection with the FIP-bias values.